Demand across China’s AI hardware supply chain is expanding beyond GPUs, with growing focus on printed circuit boards (PCB), optical interconnects, liquid cooling, and silicon carbide (SiC) substrates, according to UBS research based on company discussions.
Manufacturers are outlining production and investment plans through 2028, signaling a broader build-out of infrastructure needed to support large-scale AI deployment.
GPU hardware roadmap and execution risks
Biren Technology plans to launch its BR20X GPU in the second half of 2026, with next-generation BR30X and BR31X models targeted for commercialization in 2028. These chips are expected to deliver improvements in compute power, memory, and bandwidth.
However, challenges tied to advanced process nodes and software ecosystem development could delay timelines, underscoring ongoing execution risks in China’s domestic GPU ambitions.
PCB expansion targets AI-driven growth
Shenghong Technology said it will cap fixed asset investment at 18 billion yuan in 2026, focusing on its Huizhou production facilities. The company expects AI-related revenue to rise from under 50% to 60–70% of total income.
Management is aiming to secure about half of PCB orders tied to the Rubin computing platform, though public confirmation of these projections is still pending. Material selection for the Rubin Ultra orthogonal backplane remains under evaluation, with combinations involving Q-glass and PTFE still being tested.
Optical interconnect and cooling attract capital
Spending is also shifting toward optical networking and thermal management. Xizhi Technology, which controls roughly 88% of the independent optical interconnect solutions market, is focusing on LPO and NPO designs using 3D TSV architectures to integrate electronic ICs with silicon photonics.
These approaches are intended to reduce latency, increase bandwidth, and lower power consumption in large AI clusters.
Lens Technology has expanded into optical hardware through the acquisition of Shenzhen Tongsheng Optoelectronics, entering the hollow-core fiber cable market. The company projects optical communications revenue could reach around 10 billion yuan from 2027, while its liquid cooling segment may generate several billion yuan by fiscal 2026 if development milestones are achieved.
SiC materials see tight supply and pricing power
In power components, Tianyue Advanced Materials reported full order books for SiC products and selective price increases for smaller, urgent orders. It expects 8-inch SiC wafers to account for 50% of shipments by 2026, supporting margin expansion.
Data from Fuji Keizai shows the company held a 51.3% global market share in 8-inch SiC substrates in 2025, ranking first worldwide. Demand from electric vehicles and renewable energy is now extending into AI servers and data centers, where SiC is used in high-efficiency power systems such as HVDC and solid-state transformers.
Future pricing will depend on how quickly 8-inch capacity ramps up relative to downstream demand. Faster supply growth could limit price gains, while continued data center expansion may sustain demand.
Infrastructure build-out reshapes broader markets
UBS findings indicate that China’s AI push is spreading across the full hardware stack, from compute chips to connectivity and power systems. While companies have set clear targets, outcomes remain dependent on manufacturing yields, certification timelines, and export controls.
This large-scale investment in physical AI infrastructure is also influencing adjacent markets. Global data center spending is projected to approach $2.9 trillion by 2028, reflecting sustained demand for compute capacity.
For traders, the shift toward tangible infrastructure highlights a move away from purely speculative narratives toward measurable utility, particularly in areas tied to compute, storage, and network capacity. Decentralized infrastructure models have gained traction alongside this trend, with some networks reporting strong growth in adoption and revenue tied to real-world usage.
The expansion of AI hardware, from advanced semiconductors to optical systems, points to a deeper and more durable foundation for next-generation technologies, with market direction increasingly shaped by those positioned to benefit from this infrastructure build-out.
Explore how AI is transforming trading: discover Toobit’s powerful AI copy trading tools for data-driven crypto strategies.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
